I would be happier if it really could search and pick every unicode character. I would love to be able to nab "not equals", "approximately equals", "supsercript 2", etc.
The Unix/Linux desktop world has had compose keys for multiple decades now for that exact sort of thing. <Compose> + <=> + </> = ≠, <Compose> + <~> + <~> = ≈, <Compose> + <^> + <2> = ², etc. It's how I get em—dashes and “smart quotes” and such into whatever I type.
Only downside is that defining custom sequences is… less than intuitive.
Only downside is that defining custom sequences is… less than intuitive.